Project 04 · Case study
Rumor Discovery - Event Recommendation Engine
A sophisticated event recommendation system using weighted hybrid scoring with 5D vector embeddings, role aliasing, and advanced mathematical algorithms.
Problem
Event discovery lacks personalized mathematical precision
Solution
Vector mathematics and weighted scoring algorithms
Impact
Precise event-user matching using mathematical modeling
Users
25 user personas across 6 industries with 40 curated events
More detail
An advanced event recommendation engine that uses 5D vector embeddings, cosine similarity, and weighted hybrid scoring to match users with relevant events. Implements a scoring formula that combines vector similarity, audience fit, history, and location weighting. Features intelligent role aliasing (CEO↔Founder, VC↔Investor), pre-computed explanations for 1,000 user-event combinations, and multi-provider LLM integration. Built with Next.js and TypeScript, with test coverage including cosine similarity validation, score distribution analysis, and pipeline integration testing.